Most scientific disciplines consider anything over a 5% error rate to be a significant error rate. In previous posting some have considered the present level of certainty listed in 921, of probable as being an appropriate level. As defined, “Probable” is considered “more likely than not” or “greater than 50%”. This would give the possibility of an error rate of 49%. The National Fire Protection Association reported the average number of structure fires, between 1999 and 2002, was 517,000 per year. If we say all those fires were properly investigated and there was an acceptable error rate of 5% that would mean that there were103,400 fires where the cause was not correctly identified during that four year period. Now if we expand this using a 49% error rate then the number of fires where the cause was not properly identified was 1,013,320 over the same four year period.

Do we, as a profession, consider the 49%, or anything close, an acceptable error rate for determining the cause of a fire or is this the best we can expect as to accuracy.